Journal of Experimental Social Psychology 45 (2009) 320–326
Contents lists available at ScienceDirect
Journal of Experimental Social Psychology
journal homepage: www.elsevier.com/locate/jesp
The hidden persuaders break into the tired brain q
Christina Bermeitinger a,*, Ruben Goelz a, Nadine Johr a, Manfred Neumann a,
Ullrich K.H. Ecker b, Robert Doerr a
a
b
Saarland University, Department of Psychology, Campus A2 4, D – 66123 Saarbruecken, Germany
School of Psychology, University of Western Australia, Crawley, Australia
a r t i c l e
i n f o
Article history:
Received 30 July 2007
Revised 22 September 2008
Available online 10 October 2008
Keywords:
Subliminal priming
Subliminal persuasion
Advertisement
Subliminal advertising
Need-related subliminal stimuli
Level of tiredness
Motivation for concentration enhancement
a b s t r a c t
There is a long-lasting debate on whether subliminal advertising actually works. In this context there are
some studies suggesting that subjects’ motivation is a crucial point. Karremans et al. [Karremans, J. C.,
Stroebe, W., & Claus, J. (2006). Beyond Vicary’s fantasies: The impact of subliminal priming and brand
choice. Journal of Experimental Social Psychology, 42, 792–798] showed that subjects were influenced in
their intention to drink a specific brand of soft drink by a subliminally presented brand prime, but only
if they were thirsty. In the present study, we adapted their paradigm to the concept of ‘concentration’ and
embedded the subliminal presentation of a brand logo into a computer game. Actual subsequent consumption of dextrose pills (of the presented or a not presented brand) was measured dependent on
the level of participants’ tiredness and the subliminally presented logo. We found the same pattern as
Karremans et al. (2006): only tired participants consumed more of the subliminally presented than the
not presented brand. Therefore, the findings confirm that subjects are influenced by subliminally presented stimuli if these stimuli are need-related and if subjects are in the matching motivational state.
Ó 2008 Elsevier Inc. All rights reserved.
Introduction
Recently, Karremans, Stroebe, and Claus (2006) revived the debate
about subliminal persuasion. They connected the subliminal priming
of a brand name and the choice of this brand over control brands with
participants’ goals or needs: only thirsty participants (who were
thought to have the specific goal to quench their thirst) chose the
primed drink over the others more often. Considering the history of
findings around subliminal persuasion, to replicate and even extend
or generalize the found effect seems both challenging and vital.
Indeed, there is a long-lasting debate on the possibilities and
boundaries of subliminal advertising (for reviews, see Dijksterhuis,
Aarts, & Smith, 2005; Theus, 1994). The well known story of the
marketing expert James Vicary represents a milestone of this debate. He claimed that he had increased the sales of popcorn and
Coca Cola by 57.7% and 18.1% after the subliminal presentation
of the slogans ‘‘Drink Coca Cola” and ‘‘Eat Popcorn” in a movie
(e.g., Brand, 1978; Henderson, 1957). By 1962 at the latest, this
was exposed as a publicity hoax. Nevertheless, till today the story
is living as a modern legend (e.g., Pratkanis, 1992; Rogers, 1993;
Rogers & Smith, 1993).
q
The research reported in this article was supported by intact GmbH (Hessenweg
10, D – 48157 Muenster, http://www.intact-ms.de) who sponsored the dextrose
pills. We thank Dirk Wentura, Christian Frings, Andrea Paulus, Michaela Rohr, Joel
Cooper and one anonymous reviewer for helpful and detailed comments on an
earlier version of this paper.
* Corresponding author. Fax: +49 681 302 4049.
E-mail address: [email protected] (C. Bermeitinger).
0022-1031/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jesp.2008.10.001
Karremans et al. (2006) showed that some of Vicary’s fantasies
are maybe actually not so farfetched. In their first experiment, Karremans et al. (2006) presented a subliminal stimulus – for the
experimental group the brand name ‘‘Lipton Ice”, for the control
group the neutral string ‘‘Npeic Tol” – 25 times within a visual
detection task. Afterwards, participants had to make a virtual
choice between Lipton Ice and Spa Rood, a common mineral water.
Then, they answered questions about their intention to drink Lipton Ice, Coca Cola, and Spa Rood. Finally, the level of thirstiness
was measured with two further items. There was a positive association between the amount of thirst and the likelihood to choose
Lipton Ice (compared to Spa Rood) and between the amount of
thirst and the intention to drink Lipton Ice only in the Lipton Ice
prime condition, not in the ‘‘Npeic Tol” prime condition. In a second experiment, the authors manipulated the amount of thirst
with a salty sweet (to increase feelings of thirst) for half of the participants. Afterwards, subjects ran through exactly the same course
as in the first experiment. The main findings could be replicated:
the subliminal presentation of a drink’s name increased the
choice-probability for this drink and the intention to drink this
beverage only for thirsty individuals.
The study is perfectly in line with assumptions (and research
which confirms these assumptions) that one can be subliminally
influenced only if one is in a corresponding state with a selective
vigilance (see Bruner & Postman, 1947); that is, if the prime is in
relation to one’s current goals or needs (e.g., Brand, 1978; Strahan,
Spencer, & Zanna, 2002; Strahan, Spencer, & Zanna, 2005; and even
Vicary himself assumed that in 1958, see Rogers, 1993).
C. Bermeitinger et al. / Journal of Experimental Social Psychology 45 (2009) 320–326
The results of Karremans et al. (2006) represent a serious example of
subliminal manipulation of choice behavior with scientific means after
a long period with hundreds of mass media papers that mostly pursued
this question only insufficiently (e.g., Pratkanis & Greenwald, 1988).
But one could argue that the study is only one more study amongst
the hundred others which were – with random assignment – sometimes able to show subliminal persuasion and sometimes not. Brannon
and Brock (1994) described the problem around the lacking reproduction of a subliminal persuasion effect, fittingly, as a script for a drama
with three acts: first, observation of an effect in the field; second, demonstration of the effect in a laboratory analogue; third, subsequent failure to reproduce the effect. Therefore, the vast majority of the
psychological community concluded that subliminal persuasion does
not and could not work (for example Broyles, 2006; Dijksterhuis
et al., 2005; George & Jennings, 1975; Moore, 1988; Pratkanis & Aronson, 1992; Pratkanis & Greenwald, 1988; Trappey, 1996).
Given the lack of supporting evidence, a replication of Karremans
et al. (2006) findings is urgently needed. Besides the importance to
replicate their effect, however, the study leaves open questions
and, at one point, shares a common problem that many previous
studies have been criticized for; that is, the test of stimuli’s subliminality (e.g., Beatty & Hawkins, 1989; Brand, 1978; Brannon & Brock,
1994; Merikle, Smilek, & Eastwood, 2001; Saegert, 1979). This seems
to be especially important considering the fact that prime duration
in their study was quite long (23 ms). There are a lot of studies in
which at least some participants showed high performances in direct prime detection tests with even shorter presentation times of
17 ms (e.g., Greenwald, Draine, & Abrams, 1996).
The present research
In the present study, we wanted to take up the experiments
from Karremans et al. (2006) and conceptually replicate their findings. Additionally, we wanted to expand their results; in particular,
we measured the primes’ subliminality in the actual prime study’s
participants and we measured actual consumption behavior rather
than just the intention to behave.
First of all, we chose a need other than thirst to go one step beyond
simple physiological needs. We decided to use the concept of ‘concentration’. Tired persons need to put more effort into a task in an
achievement situation than fit persons. Therefore, tired persons
should have the motivation to enhance their concentration. Of
course, the first wish of a tired person would be to take a rest or sleep
for a while. But most often (in the lab and in everyday life as well) this
possibility is no real alternative. In the lab, participants give their
consent to concentrate and do the best they can. And in everyday life,
you only have to imagine a night-time car journey, during which it is
vital to focus on the task and mobilize your reserve resources despite
fatigue. We call this motivation the ‘motivation for concentration
enhancement’ (MfCE) and we assume that tired participants in particular have this need. This assumption was tested and corroborated
with 25 subjects who did not participate in the main experiment.1 As
1
We tested the relation of the level of tiredness and MfCE. Both were continuous
variables based on participants’ self-ratings. Materials and the procedure were similar
to the main experiment but without the direct test of discrimination performance
concerning the subliminal stimuli and without different choices of primed or nonprimed pills: the dextrose pills were offered in only one bowl which was not labeled
with any logo, and the bowl contained 20 dextrose pills. Subjects played the computer
game and were then given 28 questions mainly targeting their effort or strategies
during the game and how supporting the consumption of dextrose was. Among these,
the critical question appeared: ‘‘How strong was your need to enhance your level of
concentration before or during the game?” [2 = very, 2 = not at all]. Overall,
participants had no significant motivation for concentration enhancement, M = 0.28
(SE = 0.25), t(24) = 1.13, p = .27. Yet, most importantly, the correlation between the
level of tiredness and MfCE was significant, r = .52, p < .01; the more tired participants
were, the higher was their need to enhance their concentration level. Thus, the result
is perfectly in line with our prediction.
321
an equivalent to the drinks in the Karremans et al. (2006) study, we
chose dextrose pills – a popular means of concentration enhancement
in achievement situations.
Second, we did not measure participants’ intention to consume a
product, but their actual consumption (of a primed and a nonprimed ‘product’) within the course of the experiment. We consider this a more sensitive and valid measure than the virtual
choice of one of two alternatives (see also literature on the intention-behavior gap; for an example in a purchasing context, see
Miniard, Obermiller, & Page, 1983). First, real consumption behavior is probably less influenced by strategical thinking than virtual
choice behavior. Second, we can analyze the behavior of one person regarding both primed and non-primed ‘brands’. And finally,
it allows us to measure behavior over a longer period of time,
and also after participants have made some experience with the
different ‘products’.
A third difference is that we did not use common brands.
The two logos we used were designed especially for this study.
Additionally, one logo was subliminally presented to one half of
the subjects, the other logo was subliminally presented to the
other half of the subjects. This counterbalancing allowed us to
compare consumption behavior between both ‘brands’ instead
of comparing the subliminally presented brand and a further
control brand (as in Karremans et al., 2006). It has been discussed that influencing someone’s consumer behavior may be
simpler when concerned with neutral stimuli in comparison to
already familiar brands (e.g., Dijksterhuis et al., 2005). Nevertheless, we chose this procedure because our interest was the
manipulation of consumer behavior in terms of a relative comparison of two (novel) brands, not regarding a change of an a
priori preference.
Fourth, after the main experiment and for each participant,
we ran a direct test to objectively assess discrimination performance regarding the two logos, as recommended by Merikle
et al. (2001). Karremans et al. (2006) ran a direct test with separate participants, asking them to report the subliminally presented word. Our approach was to ask subjects to make
several two-alternative forced choices – select the logo that
was subliminally presented earlier –, which we consider a simpler and thus more sensitive task. Hence, if a participant successfully differentiated between the two logos (i.e., detected
the previously presented logo), we concluded that behavior
may have been based on some supraliminal details. Such an
individual measure of discrimination performance also allowed
us to check whether an effect hinges on participants with good
or bad performance.
Last but not least, we embedded the subliminal presentation in
a computer game, which constitutes a less artificial context than
those used in most of the previous studies.
Methods
Participants
The sample consisted of 64 (18 male, 46 female) participants
from Saarland University (Md = 21 years, ranging from 19 to 49).
They received experimental credits for their participation and
had the chance to win vouchers for the movie theater. All participants had normal or corrected to normal vision. Three subjects
were excluded from the analysis because of very low scores
(below 3 points) on the Inventar Komplexer Aufmerksamkeit
(INKA, inventory for complex attention; Heyde, 2000), testing
concentration performance (see below); these subjects likely
did not understand instructions or were not motivated to follow
them.
322
C. Bermeitinger et al. / Journal of Experimental Social Psychology 45 (2009) 320–326
Fig. 1. Black-and-white versions of the stimuli designed for the subliminal presentation. (a) logo A, (b) logo B, (c) mask created from an equal number of parts of both logos.
Design
Procedure
The design comprised factors ‘subliminal prime’ (logo A vs. logo
B) and ‘level of tiredness’ (continuous variable based on participants’ self-rating). Each participant was shown either logo A or
logo B. The number of consumed dextrose pills from product A
and product B (‘consumption’) served as the dependent variables.
Participants were tested in groups of up to four people. The
experiment was run on IBM-compatible PCs using E-Prime software and 17’’ CRT-monitors with a refresh rate of 100 Hz. Unless
otherwise noted, all instructions and tasks were given on the computer screen.
In the first phase, subjects gave their consent to participate voluntarily and with their full engagement. Participants were told that
the goal of the experiment was to test if the individual self-determined consumption of dextrose influenced concentration and performance in a computer game. They were told that we had two
different sponsors for the dextrose pills (which nevertheless had
similar taste and composition), such that they had two different
products to choose from. Subjects were additionally motivated
with the chance to win vouchers for the movie theater, which were
given to the participant with the highest score within their group.
In the second phase, participants were requested to answer
some questions about their actual state, their leisure time, and
their consumption of various substances (e.g., coffee, tea, or energy
drinks) in concentration-demanding situations. Within these questions, participants’ tiredness was measured with two five-point
items: ‘‘Today I am . . .” [2 = well rested, 2 = not well rested], ‘‘At
the moment I am feeling . . .” [2 = fit, 2 = tired]. The scores on
these items were averaged to create an index of tiredness.
In the third phase, participants were acquainted with the computer game with appropriate instructions and a two-minute practice block. This practice block was exactly the same as the
following game blocks within the fifth phase (see below), except
that the score was not included into the overall score of the participants. Within each block (i.e., during the ongoing game), there
were three sequences. In the first sequence A, which lasted 30 s,
the word ‘‘Konzentration” (concentration; 5.5 cm in width,
0.5 cm in height) was subliminally flashed on the screen in red to
enhance or trigger subjects’ motivation for concentration. The
word was presented at the same position as the subliminally presented logo; presentation time was 10 ms, with an inter-trial interval (ITI) of 3000 ms. During the second sequence B (30 s), the word
‘‘Konzentration” and the subliminally presented logo were presented alternately: Presentation started with the logo (i.e., the
prime; 10 ms), which was replaced by the mask (10 ms); after an
ITI of 3000 ms, ‘‘Konzentration” was presented for 10 ms, and so
on. In the last sequence C (60 s), only the subliminal prime
(10 ms) followed by the mask (10 ms) were presented consecutively with a 3000 ms ITI.
In the fourth phase, participants received the paper and pencil
version of the INKA (Heyde, 2000), a test of concentration and
attention (i.e., the quality of information processing).2 The test
Materials
We designed 10 colored logos (each with a graphical background
and the writing ‘Dextro’) that were tested in a pretest with 39 subjects who did not take part in the main experiment. Two logos (for
black-and-white versions see Fig. 1) were selected as subliminal
stimuli because of comparable preferences in the pretest. Both logos
could be imagined as logos for a dextrose brand and they were chosen equally often as the preferred logo. The logos did not differ with
regard to their size (7.9 cm in width, 5.8 cm in height) and their mean
luminance. The dominant colors of both logos were black, red, white,
and yellow. A mask was created by decomposing each of the two logos into 40 parts. Then, 20 parts of each logo, respectively, were newly put together (ensuring that darker parts were predominantly used
for outer areas of the image, as in the original logos; for a black-andwhite version see Fig. 1c). Dextrose pills with a diameter of 2 cm, a
weight of 2 g, and a slight citric taste served as products that could
be consumed during breaks. They were offered in two small bowls
positioned side-by-side on the left-hand side of the monitor; the
bowls were covered until the first consumption break. Each bowl
contained 10 pills; one of the logos (defining product A and product
B, respectively) was pasted onto each bowl; the positioning of bowls
(i.e., the assignment of logos to left/right bowls) was counterbalanced across subjects.
For the presentation of the subliminal logos, we designed a simple two-dimensional jump-and-reach computer game in E-Prime
(version 1.1, SP 3, Psychology Software Tools, Inc., Pittsburgh, PA,
USA). In this game, a green avatar manikin (3 cm in height) walked
on a gray wall (which occupied the lower 6 cm of the screen) and
jumped when the up-arrow-key was pressed. By doing so, the manikin could collect bundles of banknotes that flew into the screen from
the right-hand side. For each collected bundle, one point was added
to the playing score (which was shown in the right-upper corner).
Additionally, the manikin had to jump over campfires that also
emerged on the right-hand side. Two points were taken off the score
if the manikin did not succeed in doing so. To make it more difficult,
the speed of the incoming objects was increased with increasing
duration of the game. The background consisted of flickering flames
to conceal the subliminal stimulus presentations.
The logos and the mask were presented to the upper right of the
manikin. This position was chosen because it was the place most
likely attended by participants because the game objects (i.e., notes
and fires) appeared on the right-hand side.
2
This test requires participants to scan rows of consonants for specified consonants
and to transform these into assigned letters via a transformation index.
C. Bermeitinger et al. / Journal of Experimental Social Psychology 45 (2009) 320–326
was in line with the cover story and was introduced to further focus
subjects on the need for concentration. Instructions were given verbally and on the test sheets. The test lasted about 15 min.
The fifth phase comprised the main part of the experiment.
There were four game blocks of 2 min each. The procedure was
the same as in the practice block. Altogether, in each of the four
blocks, the word ‘‘Konzentration” was presented 14 times and
the subliminal prime was presented 24 times. Between the blocks,
there were consumption breaks of about 1 min each. During these
breaks, participants had the chance to consume the dextrose pills.
It was emphasized particularly that it was at participants’ discretion whether, how many, and which pills they chose to consume.
After the last game block and in the sixth experimental phase,
participants answered some questions about the game and the
dextrose brands. They were then informed about the presentation
of subliminal logos and asked if they had perceived any aspects of
the primes or masks.
This debriefing was followed by the seventh phase featuring a
direct test of discrimination performance regarding the subliminally presented logos. Each trial was set off by a key press, and then
a short sequence of the game was shown. Participants were instructed not to move the manikin but instead to attend to the position where the subliminal logo was presented. First, a green dot
was flashed for 10 ms at that position. After a 130 ms ‘‘blank” interval, the dot was repeated for a further 10 ms. Then, after another
130 ms, one of the two logos appeared for 10 ms, followed by the
mask for a further 10 ms (note that the time parameters of logo
and mask presentation were exactly the same as in the game
blocks). Finally, the two logos were presented next to each other
on a white background and subjects had to decide (using the
mouse) which logo they had just seen. This sequence was repeated
50 times, with random assignment of logos; that is, half of the trials contained the logo presented in the game, the other half contained the logo that was not presented.
Finally (phase eight), participants were fully debriefed and they
were asked to maintain confidentiality concerning the true purpose of the experiment. The score of each participant was read
from the monitor and the subject with the highest score was rewarded the cinema voucher. After the participants had left, the
number of consumed pills from each of the two bowls was
counted.
Results
Direct test of discrimination performance
For some participants, the parametric signal detection sensitivity measure d0 could not be calculated because their hit rate or false
alarm rate was either zero or one. Therefore, the non-parametric
equivalent A0 (see Pollack, 1970; Pollack & Norman, 1964) was calculated. A0 ranges from one (perfect discrimination) to zero (perfect
discrimination, but reversed keys) with A0 = .50 denoting random
responding. On average, A0 was .47 (SD = 0.16), which did not differ
significantly from random responding, t(60) = 1.34, p = .19.3
Consumption behavior
For the tired and not-tired participants (see footnote 4), the
mean number of consumed pills associated with logo A and logo B
are shown in Table 1. First, we calculated the individual difference
3
With respect to the individual 2 (presentation: logo A vs. logo B) 2 (response:
choice of logo A vs. choice of logo B) distribution, three participants had a v2 value
associated with p-values below .05 (all v2 > 4.02, all ps < .04) and another three
participants had a v2 value associated with p-values below .10 (all v2 > 2.88, all
ps < .09). Excluding these participants from the following analysis did not change the
pattern of results.
323
scores – number of consumed pills of the primed logo minus number of consumed pills of the non-primed logo. Thus, positive values
indicate more consumption of the primed product than the nonprimed product. These difference values were linearly regressed
onto the level of tiredness (from 2 = not-tired to 2 = tired). The
analysis revealed a significant positive association between tiredness and the difference of consumed pills, b = .34, t(60) = 2.79,
p < .01 (see Fig. 2) – tired subjects consumed more of the primed
product compared to the non-primed product. The constant was
significant as well, t(60) = 2.11, p < .05. The analysis yielded predicted values with a 95%-interval of confidence not including zero
for tiredness values above 0.25 (see Fig. 2). That is, even for those
participants who were neither especially tired nor especially
‘not-tired’, it can be expected that the consumption of the primed
product exceeds the consumption of the non-primed product.
The above regression analysis corresponds to an interaction between type of logo (primed vs. non-primed) and tiredness. Thus,
we calculated the simple effects of tiredness separately for primed
and non-primed logos. These analyses revealed a significant positive association between tiredness and the amount of consumed
pills only for the primed logo, b = .30, t(60) = 2.44, p < .05, not for
the non-primed logo, b = .04, t < 1 (see Fig. 3).4
Discussion
With the experiment we conceptually replicated the findings of
Karremans et al. (2006). Only participants with a specific need or
motivation chose the subliminally presented motivation-related
product more often. While Karremans et al. (2006) used the basic
need of thirst as the predictor, we transferred the problem to the
motivation for concentration enhancement, which is higher when
participants are tired – as demonstrated with an independent sample (see footnote 1). Thus, we extended Karremans et al.’s findings
to arrive at a broader understanding of motivational states.
As expected, only tired participants consumed more pills of the
primed brand than the non-primed brand; not-tired participants
consumed equal amounts of both products. Furthermore, the effect
is clearly based on the additionally consumed primed-brand pills
in the tired subsample; there was no difference between tired
and not-tired participants regarding the consumption of the nonprimed product. Thus, only the consumption of the subliminally
primed product was influenced – not the consumption of the
non-primed product. However, we do not know how the general
4
To present the more conventional results of an ANOVA and to exclude
interpretations concerning possible material differences, we additionally conducted
a repeated measures MANOVA with the factors ‘subliminal prime’ (logo A vs. logo B),
‘tiredness’ (tired vs. not-tired), and ‘bowl arrangement’ (A left and B right vs. B left
and A right), and the number of consumed pills from logo A and logo B as the
dependent variables. The majority of participants (n = 40) had negative tiredness
scores. We labeled this subsample ‘not-tired’. There were 14 participants with
positive values. These subjects were labeled as ‘tired’. Seven subjects had a tiredness
score of zero, that is, they considered themselves neither tired nor not-tired. This
subsample was labeled ‘neither-nor’. Overall, participants had a mean tiredness of
M = 0.60 (SE = 0.13), which differed significantly from zero, t(60) = 4.68, p < .001,
and indicates that they were on average rather non-tired. In the MANOVA, there was a
significant main effect of ‘tiredness’, F(1, 46) = 4.95, p < .05, gp2 = .10; tired participants consumed M = 2.31 (SE = 0.98) more dextrose pills than not-tired participants.
Further, there was a significant interaction of ‘subliminal prime’ and ‘consumption’,
F(1, 46) = 4.70, p < .05, gp2 = .09. Most importantly, these main and interaction effects
were qualified by a significant interaction of ‘subliminal prime’, ‘tiredness’, and
‘consumption’, F(1, 46) = 9.63, p < .005, gp2 = .17. As seen from Table 1 and confirming
the findings from the regression analyses, only the tired participants consumed
M = 1.7 (SE = 0.56) more pills of the primed product, t(13) = 3.07, p < .01. The
tendency of the not-tired subjects to consume M = 0.25 (SE = 0.27) more pills of the
non-primed (compared to the primed) product was not significant, t < 1. No other
main or interaction effect reached significance, all Fs < 2.9, all ps > .09. Most
importantly, there was no main effect of the subliminally presented logo, no overall
difference between the consumption of products A and B, and no effects including the
factor ‘bowl arrangement’.
324
C. Bermeitinger et al. / Journal of Experimental Social Psychology 45 (2009) 320–326
Table 1
Mean number of consumed pills from product A and product B (standard deviation in parentheses) as a function of level of participants’ tiredness (not-tired, neither-nor, tired),
subliminal prime (logo A vs. logo B) and order of bowl arrangement (A left and B right = AB vs. B left and A right = BA).
Not-tired participants
Neither-nor participants
Subliminal prime
Consumption of product A
bowl order AB
bowl order BA
Consumption of product B
bowl order AB
bowl order BA
a
b
Subliminal prime
Tired participants
Subliminal prime
logo A
logo B
logo A
logo B
logo A
logo B
2.0 (2.1)
1.2 (1.6)
1.8 (1.6)
1.8 (1.3)
0.5 (0.7)
—b
1.0 (–)a
2.8 (2.8)
4.0 (2.0)
4.0 (3.0)
1.0 (1.0)
2.4 (1.7)
1.8 (1.6)
2.1 (2.0)
2.1 (2.2)
1.0 (1.2)
1.5 (0.7)
—b
2.0 (–)a
2.5 (1.7)
2.0 (1.0)
3.0 (1.0)
3.3 (3.2)
3.6 (2.7)
There is only one participant in this condition.
There is no participant in this condition.
Fig. 2. Difference of consumed dextrose pills (consumed pills of the primed product
consumed pills of the non-primed product) as a function of the level of tiredness
ranging from 2 (not-tired) to 2 (tired) (bold line). Grey lines indicate the 95%interval of confidence; the arrow indicates the level of tiredness above which the
predicted value of the consumption difference (i.e., more consumed pills of the
primed product than the non-primed product) is significant (a = .05).
Fig. 3. Consumed dextrose pills associated with the subliminally presented logo
(i.e., the primed logo) and the not presented logo (i.e., the non-primed logo) as a
function of the level of tiredness ranging from 2 (not-tired) to 2 (tired).
consumption of dextrose was affected by the subliminal prime due
to the lack of a control group which did not receive any prime.
Therefore, it is possible that the prime per se led to an increase
of dextrose consumption. Nevertheless, for tired participants, the
subliminal prime affected the consumption of (additional) dex-
trose in a very specific manner, leading only to an increase in
primed-brand consumption. This is evidence that the prime affected participants’ behavior rather specifically. Importantly, we
were able to demonstrate the effect of subliminal priming on real
consumer behavior (and not only on the intention to consume)
in a quite realistic context.
In contrast to Karremans et al. (2006), we checked subjects’
individual prime discrimination performance directly. The low rate
of participants with above-chance performance in this discrimination task shows that the chosen presentation procedure was successful in preventing conscious awareness of the logos.
Additionally, analyses of consumption behavior yielded essentially
the same results if participants who performed above chance in
this direct test were excluded. Hence, the explanation that the effect hinges on consciously available supraliminal information is ruled out.
The prime stimuli we used were neutral and non-familiar logos
that were designed especially for use in the experiment. That may
have increased the chance of finding an effect of the subliminal
logo, because the effect of both subliminal and supraliminal advertising is greatest when participants’ attitude towards the advertised product is indifferent (e.g., Brand, 1978). Importantly,
however, logos were chosen to be equally appealing (as indicated
by pretest data), they were very similar (they displayed the same
word, they had similar colors, etc.), and each logo served as the
subliminal prime for half of the participants and as the control logo
for the other half of the participants. Consequentially, the overall
level of consumption of products A vs. B was equivalent. One can
therefore conclude that the material had equivalent a priori appeal
and that the effect did not depend on a specific logo, but on the
actually subliminally presented logo and the individual level of
tiredness. Additionally and despite claims to the contrary (Cuperfain & Clarke, 1985), we can conclude that a subliminal stimulus
does not need to have a certain a priori familiarity to be effective
in an advertising context. In a similar vein, the present study demonstrates that fairly complex stimuli can elicit subliminal manipulation of behavior (which of course requires that participants’
visual system can differentiate between stimuli, cf. Bahrami, Lavie,
& Rees, 2007). This sheds further light on the question of how complex stimuli can actually be for subliminal processing (e.g., Cooper
& Cooper, 2002; Greenwald, 1992).
The embedding of the subliminal presentation into a computer
game seems to be an interesting and potentially momentous variation to earlier studies. Our task was in no way related to the consumption product. This constitutes a less artificial context than
that used in most previous studies, which have used lexical decision tasks and the like. The fact that subjects played a computer
game means that they did not focus their attention on the position
where the subliminal stimuli would appear at all times, and also
that they were occupied with a more complex task (including,
C. Bermeitinger et al. / Journal of Experimental Social Psychology 45 (2009) 320–326
e.g., vigilant observation of the entire scene, alternate focusing and
broadening of attention, and planning of behavior) than, for example, deciding whether a string is a word or a nonword. The results
thus show that subliminal stimuli can influence behavior even in
conditions commonly considered to be suboptimal for the processing of subliminal stimuli. Hence, we brought the experimental situation closer to a more realistic advertising situation (Epley,
Savitsky, & Kachelski, 1999). In that context, a limitation of our
study lies in the circumstance that the subliminally primed product was actually present and directly accessible. Further research
could investigate whether there is transfer to situations requiring
more effort and/or (monetary) resources to obtain the actual product. Relatedly, the temporal duration of the effect needs to be
determined; in fact, the use of subliminal advertising in order to
manipulate consumer behavior will have to rely on subliminal
manipulation of memory processes and representations under
most circumstances. It should also be mentioned in this context
that subliminal advertising is illegal in some countries (including
the United Kingdom and Australia) but not others. For instance,
subliminal advertising is – strictly speaking – legal in the U.S.
(The Guardian, March 9, 2007), although the Federal Communications Commission (FCC) has the power to revoke a company’s
broadcast license (Federal Communications Commission of the
United States, 2001). Despite its legal ban, an instance of subliminal advertising with 40 ms flashes of brand logos has been reported
in Australia in 2007 (Sydney Morning Herald, February 21, 20085).
This legal state may be partially due to the failure of previous studies
(and meta-analyses) to yield convincing evidence for the effectiveness of subliminal advertising (see Introduction). Our study and that
by Karremans et al. (2006) show that this lack of evidence may be in
part due to the negligence of the motivational state of percipients.
Regarding our interpretation of the effects as motivation- or
need-related, the basic question may arise if that interpretation
is actually warranted. There are some studies showing that percipients’ passivity or relaxation favors effects of subliminal stimuli
(e.g., Brand, 1978; Fiss, 1966) and that strong needs have an ‘alerting’ effect such that all stimuli are better ingested, not just specific
need-related ones (McConnell, Cutler, & McNeill, 1959). Therefore,
it is possible that tired participants are more influenceable by all
kinds of subliminal stimuli, not just specifically motivation-related
stimuli such as brands of dextrose (note that the same argument—
that any need increases the effects of subliminal stimulation—
would hold true for the experiments of Karremans et al., 2006,
too). Although we cannot really rule out this possibility, there are
some findings suggesting that subliminal primes can act in a very
specific way. First, our own results show that it was not dextrose
consumption per se that was affected by the subliminal prime,
but the consumption of the specific product (see, e.g., Fig. 3). We
also successfully demonstrated the strong connection between
tiredness and the motivation for concentration enhancement (at
least for our test situation). Furthermore, Cooper and Cooper
(2002) reported an effect of subliminal stimuli that was specific
for one motivational state. The authors influenced subjects’ thirst
ratings by subliminally presenting the word ‘‘thirsty” and/or the
picture of Coca Cola cans, but found no increase of thirst in the control group. Additionally, they measured subjects’ hunger, mood,
and tiredness and found no influence on these variables. The
authors concluded that the motivational state changes were highly
specific to the stimuli used as subliminal primes. Given the high
specificity of consumption and the high specificity of motivational
states that can be affected by subliminal stimuli, we believe that
need-related stimuli are preferentially processed only by partici5
cf. http://business.smh.com.au/business/ten-investigated-on-splitsecond-ads20080220-1tfe.html or http://www.abc.net.au/mediawatch/transcripts/
s2082405.htm.
325
pants in the specific motivational state. Nevertheless, to buttress
our individual difference measurement, a replicatation of our study
with an experimental manipulation of tiredness seems desirable.
In sum, the present study successfully replicates and extends
Karremans et al.’s (2006) findings that subjects’ choice of a specific
product is influenced by subliminally presented information only if
they are in a motivational state congruent with the subliminally
presented stimulus. Hence, Brannon and Brock’s (1994) ‘‘accursed”
third act should be rewritten and the well known drama seems to
take a new turn.
References
Bahrami, B., Lavie, N., & Rees, G. (2007). Attentional load modulates responses of
human primary visual cortex to invisible stimuli. Current Biology, 17, 509–513.
Beatty, S. E., & Hawkins, D. I. (1989). Subliminal stimulation: Some new data and
interpretation. Journal of Advertising, 18, 4–8.
Brand, H. W. (1978). Die Legende von den ‘geheimen Verführern’: Kritische Analysen
zur unterschwelligen Wahrnehmung und Beeinflussung [The legend of the, hidden
persuaders’: Critical analyses regarding subliminal perception and influencing].
Weinheim: Beltz.
Brannon, L. A., & Brock, T. C. (1994). The subliminal persuasion controversy: Reality,
enduring fable, and Polonius’s weasel. In S. Shavitt & T. C. Brock (Eds.),
Persuasion: Psychological insights and perspectives. Boston: Allyn and Bacon.
Broyles, S. J. (2006). Subliminal advertising and the perpetual popularity of playing
to people’s paranoia. Journal of Consumer Affairs, 40, 392–406.
Bruner, J. S., & Postman, L. (1947). Tension and tension release as organizing factors
in perception. Journal of Personality, 15, 300–308.
Cooper, J., & Cooper, G. (2002). Subliminal motivation: A story revisited. Journal of
Applied Social Psychology, 32, 2213–2227.
Cuperfain, R., & Clarke, T. K. (1985). A new perspective of subliminal perception.
Journal of Advertising, 14, 36–41.
Dijksterhuis, A., Aarts, H., & Smith, P. K. (2005). The power of the subliminal: On
subliminal persuasion and other potential applications. In R. R. Hassin, J. S.
Uleman, & J. A. Bargh (Eds.), The new unconscious (pp. 77–106). New York:
Oxford University Press.
Epley, N., Savitsky, K., & Kachelski, R. A. (1999). What every skeptic should know
about subliminal persuasion. Skeptical Inquirer, 23, 40–45. 58.
Federal Communications Commission of the United States (2001). The Public and
Broadcasting. FCC record: A comprehensive compilation of decisions, reports, public
notices and other documents of the Federal Communications Commission of the
United States (v. 16, no. 18). Washington, DC: FCC.
Fiss, H. (1966). The effects of experimentally induced changes in alertness on
response to subliminal stimulation. Journal of Personality, 34, 577–595.
George, S. G., & Jennings, L. B. (1975). Effect of subliminal stimuli on consumer
behavior: Negative evidence. Perceptual and Motor Skills, 41, 847–854.
Greenwald, A. G. (1992). New look 2: Unconscious cognition reclaimed. American
Psychologist, 50, 229–238.
Greenwald, A. G., Draine, S. C., & Abrams, R. L. (1996). Three cognitive markers of
unconscious semantic activation. Science, 273, 1699–1702.
Henderson, C. (1957). A blessing or a bane? TV ads you’d see without knowing it.
Wall Street Journal, 13 [Sept., 1].
Heyde, G. (2000). Inventar komplexer Aufmerksamkeit (INKA) [Inventory of complex
attention]. Göttingen: Hogrefe.
Karremans, J. C., Stroebe, W., & Claus, J. (2006). Beyond Vicary’s fantasies: The
impact of subliminal priming and brand choice. Journal of Experimental Social
Psychology, 42, 792–798.
McConnell, J. V., Cutler, R. L., & McNeill, E. B. (1959). Subliminal stimulation: An
overview. The American Psychologist, 13, 229–242.
Merikle, P. M., Smilek, D., & Eastwood, J. D. (2001). Perception without awareness:
Perspectives from cognitive psychology. Cognition, 79, 115–134.
Miniard, P. W., Obermiller, C., & Page, T. J. (1983). A further assessment of
measurement influences on the intention-behavior relationship. Journal of
Marketing Research, 20, 206–212.
Moore, T. E. (1988). The case against subliminal manipulation. Psychology and
Marketing, 5, 297–316.
Pollack, I. (1970). A non-parametric procedure for evaluation of true and false
positives. Behavior Research Methods and Instrumentation, 2, 155–156.
Pollack, I., & Norman, D. A. (1964). A non-parametric analysis of recognition
experiments. Psychonomic Science, 1, 125–126.
Pratkanis, A. R. (1992). The cargo-cult science of subliminal persuasion. Skeptical
Inquirer, 16, 260–272.
Pratkanis, A. R., & Aronson, E. (1992). Age of propaganda: The everyday use and abuse
of persuasion. New York: W. H. Freeman and Company.
Pratkanis, A. R., & Greenwald, A. G. (1988). Recent perspectives on unconscious
processing: Still no marketing applications. Psychology & Marketing, 5,
339–355.
Rogers, S. (1993). How a publicity blitz created the myth of subliminal advertising.
Public Relations Quarterly, 37, 12–17.
Rogers, M., & Smith, K. H. (1993). Public perceptions of subliminal advertising: Why
practitioners shouldn’t ignore this issue. Journal of Advertising Research, 33,
10–18.
326
C. Bermeitinger et al. / Journal of Experimental Social Psychology 45 (2009) 320–326
Saegert, J. (1979). Another look at subliminal perception. Journal of Advertising
Research, 19, 55–57.
Strahan, E. J., Spencer, S. J., & Zanna, M. P. (2002). Subliminal priming and
persuasion: Striking while the iron is hot. Journal of Experimental Social
Psychology, 38, 556–568.
Strahan, E. J., Spencer, S. J., & Zanna, M. P. (2005). Subliminal priming and
persuasion: How motivation affects the activation of goals and the
persuasiveness of messages. In F. R. Kardes, P. M. Herr, & J. Nantel (Eds.),
Applying social cognition to consumer-focused strategy (pp. 267–280). Mahwah,
NJ: Lawrence Erlbaum Associates.
Theus, K. T. (1994). Subliminal advertising and the psychology of processing
unconscious stimuli: A review of research. Psychology & Marketing, 11, 271–290.
Trappey, C. (1996). A meta-analysis of consumer choice and subliminal advertising.
Psychology & Marketing, 13, 517–530.